File size: 3,829 Bytes
8bfe080 6b9e948 528181a 6b9e948 528181a 6b9e948 2584e80 8bfe080 ffa330f 8bfe080 8528585 8bfe080 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 |
---
license: mit
base_model: runwayml/stable-diffusion-v1-5
tags:
- stable-diffusion
- diffusion
- distillation
- flow-matching
- geometric-deep-learning
- research
library_name: diffusers
pipeline_tag: text-to-image
---
# Why do I hear boss music?
## 10000 steps
Currently retraining the scale, but it was trained with many raw unscaled latents and it makes the default output hazy.
Use this to correctly orient the output to the correct VAE scale.
## Shift 2 is the training target
Higher or lower may yield different results.
## use this
a castle at sunset
a mountain view with a beautiful landscape
a woman sitting on the bus
a carrot on a cake
a refrigerator to the left of a table
a mad scientist's laboratory with strange gagets and mechanisms
steampunk goku
a man standing on top of a table in the middle of a room full of curtains.
## 5000 steps
a mad scientists laboratory
## 4000 steps
Utilizing this synthesized image set here:
https://huggingface.co/datasets/AbstractPhil/sd15-latent-distillation-500k
As of typing this, the 500k isn't finished synthesizing. It's at around 200k, which should be more than enough to get a baseline.
At 4000 steps the new flow matching trainer is already manifesting results.
Within 4000 steps at batch 16 the pretrained flow matching SD1.5 model is already building convergence.
This model was the sd15-flow-matching-try2 aka Lune variation, and I can say for certain she is most definitely not burned.
The trainer is in the files.
|